Effective Movement Classification for Context Awareness in Medical Applications Networking

Abstract

Today new medical applications evolve from large stationary devices to small and smart mobile systems that will enable e.g. more efficient post operative health care. These mobile systems that benefit from ongoing miniaturization and energy savings in hardware will allow continuous monitoring of patients accompanying and supporting therapy and detect emergency situations. Additionally to vital data, the physiological load or the context of patients are important to analyze and understand recorded data of mobile patients. Current approaches for movement classification aim to detect very specific movement patterns and are dependent on precise sensor placements and are thus not suited for everyday usage. Therefore, we developed a movement detection and classification algorithm that can be easily integrated in existing embedded devices. Using data from a single accelerometer embedded into the device, the algorithm can classify between different movement patterns - the "context" - of the monitored person. We will describe the hardware and the algorithm and will provide first evaluation results demonstrating the effectiveness of this approach for providing context awareness in mobile medical applications in real-time.